5800X Boost set to 5050Mhz all cores with AMD Curve Optimizer

[ubasti]

Yeah this seems counter intuitive, but the logic from what I've gathered is windows will hit your 2 best cores the hardest. The others are more likely to boost as part of all core to a lower level in effective clock, and lower curve helps get these a bit higher, but because these won't be pegged to sit at 5GHz effective, stability should be okay. While your weaker cores will record clock highs above 5GHz, they are not likely to go to that level in effective clocks at all, let alone for extended periods. When I tried any lower than -5 on my 2 best, instant crash in R20, even launching it, so this seems to align with the theory.

I think you're right and further experimentation I have done seems to support this. Initially I took the AMD guidance on their presentation a little too literally perhaps, set the PBO limit to the maximum of 200mhz, which on a 5900x would be 5150mhz. Then when I introduced negative values on the curve optimiser for my best 2 cores, this introduced instability at anything over -8. However, I've since realised this is likely due to the cure then allowing the cores to boost beyond where they're stable, causing crashes. When I capped the PBO frequency at +100 (5050) I've then been able to take the 2 best cores to -12 stable, with cinebench results peaking at -10 then tapering off at -12.

I have then also been able to introduce -15 curve optimisation on the remaining ten cores, which has benefitted multi core results. I am now hitting the sort of cinebench R20 scores most reviewers were getting (8400-8500) but my machine was falling short of. I'm starting to appreciate there is nothing that 'wrong' with the cpu, as most reviews used a very small number of AMD-sanctioned boards to hit those numbers, and the X570 Tomahawk seemed to be running out of juice on all-core R20 using a 5900x under default settings even when temperatures were under control.

 

[Guest2]

How do you know its too much? Just a value looking scary?

Its too much when temps are pushed to 90c + on an AMD recommended Ryzen 5000 cooler and AIDA64 shows stability errors.
On 1.5V I got high temps and stability errors. 1.275V all core i get no stability errors (2 hours AIDA stability testing so far) and max temp is 76c.

 

[Dean Guile]

To be fair AMD Robert has said many times due to the amount of people worryingly posting about it, that Zen 3 voltages up to 1.5v is not a problem and were built for it, and temps exceeding 80c is now the way these new cpus work, trying to get performance what it can up to around 80c, rather than lowering temps and being happy with lower speeds, just trying to convince myself this is the case, i'm slowly coming round.

Which makes sense them telling us to get top market coolers, as usually when we do this we just see lower temps and can work up, but now the cpu is basically doing that for us all the way up to 80c, so basically better cooler = faster cpu, which is what we all been trying to do with overclocking in the past, but basically zen 3 has done it all for us, and built all that into the algorithm so all we got to do for overclocking now, is buy a good/better cooler then the chip does the rest all within it's own safety parameters.

Thinking about it now after writing this, Zen 3 is bloody amazing!!!!

 

[LtMatt]

To be fair AMD Robert has said many times due to the amount of people worryingly posting about it, that Zen 3 voltages up to 1.5v is not a problem and were built for it, and temps exceeding 80c is now the way these new cpus work, trying to get performance what it can up to around 80c, rather than lowering temps and being happy with lower speeds.

This.

 

[xAliasx]

Its too much when temps are pushed to 90c + on an AMD recommended Ryzen 5000 cooler and AIDA64 shows stability errors.
On 1.5V I got high temps and stability errors. 1.275V all core i get no stability errors (2 hours AIDA stability testing so far) and max temp is 76c.

If I understand correctly, the 1.5V is only when single cores are boosting and not when all cores are in play. If you do a all core stress test, you can see that the voltage drops significantly

 

[Dean Guile]

If I understand correctly, the 1.5V is only when single cores are boosting and not when all cores are in play. If you do a all core stress test, you can see that the voltage drops significantly

That's right, as AMD Robert also said, 1.5v is only briefly and per core.

 

[Guest2]

If I understand correctly, the 1.5V is only when single cores are boosting and not when all cores are in play. If you do a all core stress test, you can see that the voltage drops significantly

Correct, 1.375 seems too high for all core which is about where it settles for 4.6Ghz (at least on my 5800X & motherboard)

 

[deagle]

That's right, as AMD Robert also said, 1.5v is only briefly and per core.

What about 1.4-1.45v which is pretty much sustained over a medium load session (gaming)? I hear you but my cynical side says AMD is forced to ramp up volts to match Intel in single threaded/gaming workloads and being a little hand wavey with the risks. Degredation is almost impossible to prove and firmware could start scaling down over time without 99% of users noticing.

 

[Dean Guile]

I was simiar to this thinking at the start of zen 1 with my 2700x, but lo and behold it was absolutely fine, and that cpu is still going strong in my sons system now up to near 1.5v sometimes.
Degradation is also something I don't get with people, average lifespan of an average cpu is 10 years, and that was over a decade ago, it's probably longer now, but for arguments sake lets take 10 years, do you really think you will have this same cpu in 2031???? (just for perspective 10 years ago what cpu did you have in 2010, is it even relevant now? even slightly?) say you do something so drastic that it kills the lifespan by a whole 50%!!! that's 5 years now, do you still even see yourself with this same cpu in 2026????

I would trade 50% lifespan for 10%+ performance gain all day long!

 

[Dean Guile]

Also another argument I always had with my R8 and R35, I never ever garaged them for the winter like some garage queens, and people always said to me, your killing the resale value, I mean wtf! Why on earth would I care about the next owner, why would I restrict the enjoyment of my current ownership just to have better resale for the new owner, worry about yourself, not where or who it's going to next. lifes too short!"

 

[thebandit]

There was a video linked some pages back from AMD I think in this thread with the explanation of what is happening inside the CPU. Its just playing power, volts and temp off against each other in different work loads to achieve the best performance. The voltage on its own isn't the issue, its the power draw which generates the heat (combined voltage and current of all cores) There are maximums for each parameter but the all three wont max out together, 1 will always be sacrificed for the benefit of the other 2 depending on the workload was the way it was described. Getting a better cooler just raises total envelope so you can push higher but the same rules apply.

 

[Dean Guile]

I think the problem is, so many people got used to intel and temps in the 20s, and highs of 60s, with crazy coolers, that it's hard to get used to chips of Zen 3 pushing right to the edge, but at the same time well within boundaries of any sort of risk, I've had a 2700x, then 3900xt, and even I;m still coming to terms with this 5900x and the way it works.

Also further to my earlier comment about caring too much about resale of lifespan, 10 years ago this was my system, that's a core 2 duo 6600 in there, it's now 10 years later, puts it into perespective a bit doesn't it, when someone says this or that gives faster degradation, core2duo is not even slightly cared for nowadays and is probably melted and been recycled already, do I care? no, did I enjoy it, hell yes, overclocked the shiz out of it! (GPU is 8800gtx)

 

[schoenzy]

I think the problem is, so many people got used to intel and temps in the 20s, and highs of 60s, with crazy coolers, that it's hard to get used to chips of Zen 3 pushing right to the edge, but at the same time well within boundaries of any sort of risk, I've had a 2700x, then 3900xt, and even I;m still coming to terms with this 5900x and the way it works.

Also further to my earlier comment about caring too much about resale of lifespan, 10 years ago this was my system, that's a core 2 duo 6600 in there, it's now 10 years later, puts it into perespective a bit doesn't it, when someone says this or that gives faster degradation, core2duo is not even slightly cared for nowadays and is probably melted and been recycled already, do I care? no, did I enjoy it, hell yes, overclocked the shiz out of it! (GPU is 8800gtx)

I had a Q6600 and 8800GTS 512. Classic era. I went 2500k after that, then 4690k platform used off a buddy, upgraded to a used 4790k to prolong further 2 years ago. Just jumped to 5800x.

When I built my first pc with the Q6600, I also built one for my Opa. He's all emotional about it and has kept that thing on life support because I built it.. so he's one of the few out there who has a 10+yr cpu lol

 

[Dean Guile]

I know tell me about it, that 8800gtx was incredible for it's time, and it had really cool graphics on it, was an XFX one, who always put graphics on the actual cooler, and their cases were all a big X, good ol days.

 

[schoenzy]

I know tell me about it, that 8800gtx was incredible for it's time, and it had really cool graphics on it, was an XFX one, who always put graphics on the actual cooler, and their cases were all a big X, good ol days.

Yeah, that's when cooler graphics took off, so cool. This is what I had

Spoiler

Sorry for the off topic lolll

 

[ubasti]

for arguments sake lets take 10 years, do you really think you will have this same cpu in 2031???? (just for perspective 10 years ago what cpu did you have in 2010, is it even relevant now? even slightly?) say you do something so drastic that it kills the lifespan by a whole 50%!!! that's 5 years now, do you still even see yourself with this same cpu in 2026????

I would trade 50% lifespan for 10%+ performance gain all day long!

Depends, if game architecture goes the way indicated by titles such as Rise of the Tomb Raider and Horizon Zero Dawn where the 5900x beats the 5600x significantly in a non-CPU bound conditions, then certainly 5 years and maybe more.

A good case study is the Intel 2500k I bought in 2011, I still use it for some stuff. It was also overclocked for 5 years and some people have had theirs overclocked for 9 years! What really made me move off it as my main machine was not so much the performance of the CPU but the chipset / motherboard feature set: Slow DDR3 ram, some SATA connectors had failed, no M.2 or USB3, it's the stuff surrounding a high-performing CPU that falls behind before the chip itself. It could be the same with future platforms and things like DDR5, we will just have to wait and see, but I am definitely hoping to get 5+ years gaming at the high end out of this, with a replacement GPU for the 3080 along the way.

 

[Dean Guile]

Remember guys I was being conservative with 10 years, in reality the normal lifespan is a lot longer.

 

[deagle]

Honest question, is a 2500k running 1.2v on an older Intel process a fair metric of longevity vs TSMC 7nm being pushed to 1.45v? I don't need to die on this hill btw, I'm happy to throw cooling at the problem and never look at voltage again if this is truly safe

 

[gerardfraser]

What is mind boggling to me is that people in this thread are second guessing the hundreds of AMD engineers who actually make the product and have perfect understanding how the CPU works.

 

[Dean Guile]

What is mind boggling to me is that people in this thread are second guessing the hundreds of AMD engineers who actually make the product and have perfect understanding how the CPU works.

Just shows great testament to what AMD have achieved imo, even myself who's had Ryzen since 2700x, and now the 5900x, still coming to terms with the power of these little beauties.